{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:7CB3U6ZYSKBY5SWI6M5UJRN5ES","short_pith_number":"pith:7CB3U6ZY","schema_version":"1.0","canonical_sha256":"f883ba7b3892838ecac8f33b44c5bd24b7f2d742542066cd97727d9a8c6c2dc8","source":{"kind":"arxiv","id":"2606.00257","version":1},"attestation_state":"computed","paper":{"title":"ARCA: Adapter-Residual Credit Assignment When Token Signals Degenerate","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Rodney Lafuente-Mercado","submitted_at":"2026-05-29T18:42:06Z","abstract_excerpt":"Token-level credit assignment for language-model reinforcement learning is usually formulated as if the policy were fully trainable, while practical LLM-RL pipelines often rely on parameter-efficient fine-tuning, especially LoRA. We argue that this separation hides a structural failure mode. Under LoRA, the policy is restricted to a low-rank neighborhood of the reference model, so the per-token output-distribution differences used by common intrinsic credit signals, surprisal, entropy reduction, and policy divergence, can become degenerate after within-trajectory normalization, either approach"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2606.00257","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-29T18:42:06Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"acec681af515fc344a2392453ffe315d749c810879ea2fd17eccf5a621c71076","abstract_canon_sha256":"0654e981f8ae7658ea75143c9edf78584eca7ffa4ef43098c9b269af55eb98db"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T01:03:22.896243Z","signature_b64":"4oE17vJ/QFwvNFbxXazpD9aE9bXVBYYRObCwSddCN8pMeKdWBR6iUQr5gCb29FVZ8Uey3SWZrA6V93m+36mMCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f883ba7b3892838ecac8f33b44c5bd24b7f2d742542066cd97727d9a8c6c2dc8","last_reissued_at":"2026-06-02T01:03:22.895880Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T01:03:22.895880Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ARCA: Adapter-Residual Credit Assignment When Token Signals Degenerate","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Rodney Lafuente-Mercado","submitted_at":"2026-05-29T18:42:06Z","abstract_excerpt":"Token-level credit assignment for language-model reinforcement learning is usually formulated as if the policy were fully trainable, while practical LLM-RL pipelines often rely on parameter-efficient fine-tuning, especially LoRA. We argue that this separation hides a structural failure mode. Under LoRA, the policy is restricted to a low-rank neighborhood of the reference model, so the per-token output-distribution differences used by common intrinsic credit signals, surprisal, entropy reduction, and policy divergence, can become degenerate after within-trajectory normalization, either approach"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.00257","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2606.00257/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2606.00257","created_at":"2026-06-02T01:03:22.895936+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.00257v1","created_at":"2026-06-02T01:03:22.895936+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.00257","created_at":"2026-06-02T01:03:22.895936+00:00"},{"alias_kind":"pith_short_12","alias_value":"7CB3U6ZYSKBY","created_at":"2026-06-02T01:03:22.895936+00:00"},{"alias_kind":"pith_short_16","alias_value":"7CB3U6ZYSKBY5SWI","created_at":"2026-06-02T01:03:22.895936+00:00"},{"alias_kind":"pith_short_8","alias_value":"7CB3U6ZY","created_at":"2026-06-02T01:03:22.895936+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7CB3U6ZYSKBY5SWI6M5UJRN5ES","json":"https://pith.science/pith/7CB3U6ZYSKBY5SWI6M5UJRN5ES.json","graph_json":"https://pith.science/api/pith-number/7CB3U6ZYSKBY5SWI6M5UJRN5ES/graph.json","events_json":"https://pith.science/api/pith-number/7CB3U6ZYSKBY5SWI6M5UJRN5ES/events.json","paper":"https://pith.science/paper/7CB3U6ZY"},"agent_actions":{"view_html":"https://pith.science/pith/7CB3U6ZYSKBY5SWI6M5UJRN5ES","download_json":"https://pith.science/pith/7CB3U6ZYSKBY5SWI6M5UJRN5ES.json","view_paper":"https://pith.science/paper/7CB3U6ZY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.00257&json=true","fetch_graph":"https://pith.science/api/pith-number/7CB3U6ZYSKBY5SWI6M5UJRN5ES/graph.json","fetch_events":"https://pith.science/api/pith-number/7CB3U6ZYSKBY5SWI6M5UJRN5ES/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7CB3U6ZYSKBY5SWI6M5UJRN5ES/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7CB3U6ZYSKBY5SWI6M5UJRN5ES/action/storage_attestation","attest_author":"https://pith.science/pith/7CB3U6ZYSKBY5SWI6M5UJRN5ES/action/author_attestation","sign_citation":"https://pith.science/pith/7CB3U6ZYSKBY5SWI6M5UJRN5ES/action/citation_signature","submit_replication":"https://pith.science/pith/7CB3U6ZYSKBY5SWI6M5UJRN5ES/action/replication_record"}},"created_at":"2026-06-02T01:03:22.895936+00:00","updated_at":"2026-06-02T01:03:22.895936+00:00"}